Has any sense to talk about temperature distribution in a so fast process and a material with so high thermal conductivity? Perhaps in the first moment you may talk about temperature, as well as the nuclear scientists estimate the temperature of a termonuclear explosion; but other thing is the modeling of the kinetic energy diffusion throw material.

Nevertheless, the Fourier law is linear, and if you consider that in your extreme conditions an stationary process is valid in principle, then you could apply it as a work hypothesis. 

Dear  Fang Sheng Lim,

First, I am not what you mean by `temperature distribution`.  But if your goal is to estimate the thickness of the thin gold layer as a function of some deposition parameter, I think that Beer-Lambert is not the best approach.  Personally, I would characterize the samples using both transmittance (assuming that the gold layer is semitransparent) and reflectance (or better, ellipsometry).  For the analysis of the data, you should use an optical thin film model (i.e. Abeles matrix approach), keeping in mind that a thin semitransparent gold layer will have a different index of refraction (and extinction coefficient) depending on its thickness.